Benchmark

SkillDev tools

Run benchmarks and update docs/benchmarks.md. Use before releases or when comparing against other projects.

Available today. Use it from your connected AI after setup.

Connect ahel once, and every AI you use reads what you have installed.

Then ask your AI: use the Benchmark skill

What this skill tells your AI

The instructions your AI receives, as published by cniska/acolyte in .agents/skills/benchmark/SKILL.md and read by ahel’s review.

Run the benchmark script, parse its output, and update docs/benchmarks.md with fresh numbers and today's date.

Workflow

  1. Always delete /tmp/acolyte-benchmarks and reclone all repos before running. Incremental pulls frequently leave repos in a broken state with missing files.
  2. Run bun run scripts/benchmark.ts and capture the output.
  3. Parse the output — each project block lists metric labels and values. Match each value to the corresponding table cell in docs/benchmarks.md by project name and metric label.
  4. Update every metric value the script covers. Do not change narrative prose, section headings, or table structure.
  5. Update the date line at the bottom of the file to today's date in the form Updated <Day Month Year>.
  6. Commit the result with a message in the form chore(docs): update benchmarks <date>.

Rules

  • Update numbers and the date only — leave all prose and table structure intact.
  • If a metric from the script has no matching cell in the doc, skip it.
  • If a cell in the doc has no corresponding script output, leave it unchanged.

Signals

GitHub stars
25
Last commit
Sep 2026
Hacker News mentions
20
Advanced
Catalog kind
skill
Gateway key
benchmark-cniska
Source
github.com/cniska/acolyte